Analyze competition deeply

Analyze a competition deeply

Purpose

Convert scattered official materials into a verified contest brief that the idea-planning skill can use without repeating research. Analyze the contest the user intends to enter; do not decide whether they should participate.

Boundary

Own:

• official rules, eligibility, schedule, deliverables, anonymity, rights, and disqualifiers;
• rubric weights and hidden reviewer questions;
• organizer mission, current programs, terminology, and likely use context;
• official prior winners and adjacent alternatives;
• team-fit opportunities, evidence gaps, preparation priorities, and unresolved questions.

Do not:

• generate a finished idea portfolio;
• write the proposal;
• predict award probability;
• use GO, CONDITIONAL GO, or NO-GO labels;
• treat edited award summaries as original submissions.

Hand idea generation to $plan-competition-ideas. Hand draft auditing to $review-competition-proposals.

Start with a blind pass

Before research, identify 2–3 overlooked issues that could materially alter preparation, such as ambiguous eligibility, duplicate-award restrictions, data-use rights, required prototype state, or a conflict between notice and form.

If essential input is missing, ask at most three concise questions. Otherwise state assumptions and continue. Never request personal identifiers that are unnecessary for analysis.

Required workflow

1. Lock the source set

Collect the notice, attachments, application form, rubric, FAQ, correction notices, submission page, and organizer pages. Prefer official primary sources. Browse the web for current requirements, deadlines, policies, organizers, awards, or other changeable facts.

Record for every source:

• title and issuing organization;
• URL or file name;
• publication or update date;
• checked date;
• original, edited summary, news release, or secondary source;
• claim supported and limitation.

Use references/research-protocol.md for source hierarchy and conflict handling.

2. Build the requirements matrix

Extract without paraphrasing away legal meaning:

• eligibility and team limits;
• category, topic, and mandatory technology or data;
• deadline with timezone and submission channel;
• file type, naming, page/character/byte limits;
• mandatory headings, attachments, signatures, demonstrations, and links;
• anonymity, 개인정보, copyright, portrait, licensing, duplicate submission, and prior-award rules;
• evaluation stages, rubric, weights, bonus points, and disqualifiers;
• post-selection obligations, presentation, prototype, commercialization, or ownership terms.

Separate confirmed, conflict, not provided, and organizer confirmation required. Do not infer a rule from a checkbox or template layout alone.

3. Translate the rubric into reviewer questions

For each official criterion, state:

• the reviewer question;
• evidence that would answer it;
• common weak answer;
• likely document location;
• current team evidence and missing evidence.

If no official rubric exists, label the questions as inferred. Do not invent weights.

4. Analyze organizer intent

Distinguish:

• explicit contest purpose;
• organizer mission and active programs;
• target beneficiary, operator, buyer, or policy user;
• expected post-award use;
• inference from sources.

Show the reasoning chain for every inference. Do not confuse organizer publicity language with a binding requirement.

5. Study winning and adjacent cases

Use official award materials first. For each relevant case, code:

• target and concrete problem scene;
• root cause or unresolved gap;
• intervention mechanism;
• evidence and implementation state;
• execution owner and measurable effect;
• writing or visual pattern;
• weakness or missing proof in the public summary;
• transferable pattern versus contest-specific feature.

Do not copy wording. Do not call finalists winners. State when only an edited profile is available.

6. Map team fit and preparation leverage

Connect team skills, prior assets, available data, field access, partners, design, and implementation experience to rubric questions. Treat team fit as execution evidence, not biography.

List preparation leverage as:

• usable now;
• usable after verification;
• unavailable or permission-dependent.

7. Produce evidence and risk priorities

For each material issue provide:

finding → evidence → interpretation → preparation implication → verification action

Prioritize by submission blocking risk, rubric impact, evidence difficulty, and lead time. Because the user intends to enter, conclude with:

• strongest preparation advantages;
• critical unknowns;
• evidence to secure first;
• organizer questions;
• recommended idea search zones;
• next-step handoff.

Never conclude with a participation verdict.

Evidence language

Tag claims as [공식 사실], [현장 근거], [팀 분석], [검증 결과], [가정], [추론], [목표], or [계획]. Mark missing material as [출처 필요], [기관 확인 필요], [분석 필요], or [검증 필요].

Absence of search results is not proof of nonexistence. Use “확인한 공식 자료 범위에서는 찾지 못했다” with search scope and date.

Output contract

Return, at minimum:

1. executive brief;
2. overlooked issues and assumptions;
3. source register;
4. requirements and disqualification matrix;
5. rubric-to-reviewer-question map;
6. organizer intent analysis;
7. prior-winner and adjacent-case pattern table;
8. team-fit and evidence map;
9. prioritized research, organizer questions, and preparation actions;
10. handoff block using references/handoff-schema.md.

Scale detail to the contest. Explain the reason for every priority or recommendation.

References

• Read references/research-protocol.md before web research or source comparison.
• Use references/handoff-schema.md for the final artifact passed to $plan-competition-ideas.
• Use references/question-bank.md when the user asks how to provide better input.